The User-first Approach to AI Ethics: Preferences for Ethical Principles in AI Systems across Cultures and Contexts

Users prioritize privacy, fairness, and transparency, but not equally: The study found that while these three ethical principles consistently rank highest across users, preferences vary significantly—suggesting that a one-size-fits-all approach to AI ethics may not reflect what people actually want from AI systems.

Culture and context dramatically shape ethical preferences: The research conducted across four countries revealed that users' values for specific ethical principles shift depending on both their cultural background and the particular application of AI, indicating that responsible AI design must be locally informed rather than universally standardized.

A significant portion of users are "ethically disengaged": Latent class analysis identified that the largest user cohort defers ethical decision-making to regulatory bodies rather than engaging with ethical principles themselves, highlighting a potential gap between expert-driven AI ethics frameworks and public engagement.

Current AI ethics frameworks may not reflect user priorities: The paper provides empirical evidence that the ethical principles emphasized by designers and policymakers don't necessarily align with what users actually value, suggesting that operationalizing AI ethics requires direct input from the people affected by these systems.

Effective AI governance requires both regulation and user-centered design: The findings support developing robust regulatory mechanisms while simultaneously tailoring ethical implementations to specific cultural contexts and use cases, rather than relying solely on abstract moral principles.